def fit_model(n_estimators=500, learning_rate=0.26):
params = {
'n_estimators': 100,
'max_depth': 200,
'min_samples_split': 400,
'learning_rate': 0.1,
'verbose': 2,
'loss': 'ls',
'random_state': 0
}
# params = {'n_estimators': 500, 'max_depth': 10, 'min_samples_split': 2,
# 'learning_rate': 0.01, 'verbose': 1, 'loss': 'ls', 'random_state': 0}
# params = {'n_estimators': 500, 'max_depth': 4, 'min_samples_split': 2,
# 'learning_rate': 0.01, 'loss': 'ls'}
# n_estimators: 500, 'max_depth': 5, 'min_samples_split': 6,
# learning_rate: 0.26,
xgb = ensemble.GradientBoostingRegressor(**params)
# gbrt = ensemble.GradientBoostingRegressor(verbose=2)
xgb = XGBRegressor(verbose=2)
# gbrt = ensemble.GradientBoostingRegressor(loss='ls',n_estimators = 300,max_depth = 300, learning_rate = 0.1, verbose = 2, min_samples_leaf = 256, min_samples_split = 256)
fileName = 'E:\\data\\DiDiData\\data_csv\\result\\default_para_xgb_pair_result'
global x_train, y_train
# gbrt.fit(x, y)
# gbrt.fit(x_test, y_test)
xgb.fit(x_train, y_train)
save_model(fileName, xgb)
model_result = xgb.predict(x_test)
save_result(fileName, list(model_result))
mse = mean_squared_error(y_test, xgb.predict(x_test))
print("MSE: %.4f" % mse) # 输出均方误差
r2 = r2_score(y_test, model_result)
print("r^2 on test data : %f" %
r2) # R^2 拟合优度=(预测值-均值)^2之和/(真实值-均值)^2之和,越接近1越好
# return mse
# Plot training deviance
# compute test set deviance
test_score = np.zeros((params['n_estimators'], ), dtype=np.float64)
for i, y_pred in enumerate(xgb.staged_predict(x_test)):
test_score[i] = xgb.loss_(y_test, y_pred)
plt.figure(figsize=(12, 6))
plt.subplot(1, 2, 1)
plt.title('Deviance')
plt.plot(np.arange(params['n_estimators']) + 1,
xgb.train_score_,
'b-',
label='Training Set Deviance')
plt.plot(np.arange(params['n_estimators']) + 1,
test_score,
'r-',
label='Test Set Deviance')
plt.legend(loc='upper right')
plt.xlabel('Boosting Iterations')
plt.ylabel('Deviance')
# Plot feature importance
feature_importance = xgb.feature_importances_
# make importances relative to max importance
feature_importance = 100.0 * (feature_importance /
feature_importance.max())
sorted_idx = np.argsort(feature_importance)
pos = np.arange(sorted_idx.shape[0]) + .5
plt.subplot(1, 2, 2)
plt.barh(pos, feature_importance[sorted_idx], align='center')
columns = x_train.columns
plt.yticks(pos, columns[sorted_idx])
# plt.yticks(pos, boston.feature_names[sorted_idx])
plt.xlabel('Relative Importance')
plt.title('Variable Importance')
plt.show()
